NSF definition from Goodman et al (Science 2016): Reproducibility refers to the ability of a researcher to duplicate the results of a prior study using the same materials as were used by the original investigator.
Do we agree?
Perhaps we should be talking about “Methods Reproducibility”
It is not the truth, or even an approximation of truth Reproducing the same bias/mistake = same (wrong) results
If most researchers think this is important,
why are we not working reproducibly?
What are the obstacles?
Think-Pair-Share (5 min)
http://www.socrative.com 1 Student login, Room Name :AHREN
*I will add a picture for this later…
How is your work affected by procrastination? (5 min) > http://www.socrative.com 2 & 3
> Student login, Room Name :AHREN
F: Fusion - Fusing the topic with bad thoughts about failure:
“I can not do this”
“It is too hard”
“It will take me too long time”
E: Excessive Goals - We get excited and want to change everything immediately
This leads to to unrealistic expectations and a high risk of failure!
Hard to keep up long term
A: Avoidance of Discomfort - All types of change will feel uncomfortable and unfamiliar at first
Avoiding discomfort, taking the easy way out (=no change)
Change is scary!
R: Remoteness from Values - Easy to loose goal and purpose
“Why am I even doing this?”
Note: Get organised! does not help!
Note: I am not expert in all these topics
Time: 10 minutes plus time for sharing
Make a mind map to visualise the process of one of your projects.
Be as detailed as you can regarding files needed and produced. Is external data required?
How would you like to structure this project? Draw a management plan on a new piece of paper.
Include files, folder structure, project structure and external resources required
Put the two pieces of paper up for viewing by others.
Include information about the goal and reasoning for the project README
Progs
Read-only, raw data and meta data
This is an exact COPY of the data at the start of the project
Put documentation (e.g Rmarkdown, Notes etc)
Scripts, such as sbatch, bash, R scripts etc
Store software installed locally, also keep source code
Make a separate folder for each of the steps in the analysis I like to number them to get a nice order
Now what?
What is Git?
Git is a free and open source distributed version control system designed to handle everything from small to very large projects with speed and efficiency.
What is GitHub?
GitHub is a web page were git repositories can be shared. It is a essentially social platform for code. Good for most things that fit with Git.
https://jennybc.github.io/2014-05-12-ubc/ubc-r/session03_git.html
stage (add)
commit
push
pull
clone
branch
Sign up to Github https://github.com/
Create a new Github repo for this workhop
Give the repo the name of your project
Check the box initialize this repository with a README
File-> New Project -> Version Control -> Git
Paste in the git hyperlink found on the github repo page under Clone or download-> SSH
Update your project and file structure based on the discussions from your peers. Save the changes, click the Git tab on the top right window -> commit -> push
In Github repo: -> Settings -> Collaborators -> Search by email of Github username
Think-Pair-Share
http://www.socrative.com 4 Student login, Room Name :AHREN
Comments?